Demagnetization Diagnosis Of Vehicle PMSM Based On Flux Observation And Phase Current Analysis

With the increasing consumption of oil resources,the regulations of greenhouse gas emission are gradually stricter.Hybrid vehicle and EV are gradually replacing gasoline vehicles.It is of great significance to carry out research on fault diagnosis of vehicle motor system for ensuring the safe operation of the vehicle.Functional safety standards require that motors have a fault diagnosis mechanism,which can detect faults in time and take corresponding warning and protection strategies.The working scene of PMSM is complex.Under severe conditions such as high speed increase,motor overheating,and abnormal drive circuit,the motor is demagnetized and can’t output the target torque.Magnet demagnetization is divided into two cases:uniform and partial demagnetization.This paper will study the diagnosis methods of this two conditions:establish a uniform demagnetization diagnosis method based on RLS and improve the diagnosis method under dynamic conditions;Local demagnetization establishes VMD and SVM identification fault diagnosis mechanism suitable for dynamic conditions.The main contents of this article are as follows:1.Analyze the mathemati model of the motor under static three-phase and rotating coordinate system.When the permanent magnet is demagnetized,the permanent magnet torque decreases.A voltage equation considering non-integer harmonics and integer harmonic flux linkages is established,which provides a theoretical basis for the analysis of local demagnetization fault.2.Analyze the demagnetization mechanism and establish the JMAG model of the drive motor.The effects of uniform and local demagnetization on internal magnetic density distribution and no-load back EMF are analyzed.Under uniform demagnetization,the fundamental wave magnetic density and flux linkage value decrease accordingly;under local demagnetization,the k/p magnetic harmonics increase accordingly.3.For uniform demagnetization:The flux linkage identification algorithm is established based on the RLS.The simulation finds that the identification error reaches 40% when the load torque step changes.The algorithm is improved,the current change rate is introduced into the output matrix,and the threshold e is set.When the current change is greater than e,the identification algorithm considering the current change rate is used;when it is less than e,the traditional algorithm is used.Experiments show that the identification error of the improved algorithm is less than 3% when the load step changes,and it can be used for uniform demagnetization fault diagnosis under dynamic conditions.4.For local demagnetization:establish a data-driven diagnosis method based on phase current,use PSO to optimize VMD decomposition parameters,perform current VMD decomposition;select sensitive IMF for reconstruction and energy entropy calculation;use time domain features and energy entropy as fault features Perform SVM classifier training and validation,and the accuracy rate reaches 95%.The speed change affects the VMD decomposition effect.According to the order ratio tracking method,the current combined with the rotation angle signal is resampled in the angle domain,and the modal decomposition and SVM training and verification are carried out.Experiments show that the improved algorithm can complete local demagnetization fault diagnosis under dynamic conditions.